Using the null-stream of GEO 600 to veto transient events in the detector output

Abstract

A network of gravitational wave detectors is currently being commissioned around the world. Each of these detectors will search for gravitational waves from various astronomical sources. One of the main searches underway is for un-modelled, transient gravitational wave events. The nature of these signals is such that it will be difficult to distinguish them from bursts of instrumental noise that originate in or around the detector and which then couple to the main detector output. One way to deal with this is to look for events that are coincident in more than one gravitational wave detector. However, with very large event lists (potentially thousands of events per day per detector), the number of events that pass this test due to random chance can still be large. At each detector site, various methods are being developed to veto instrumental bursts from lists of candidate events from that particular detector. This reduces the size of the event lists of each detector, and hopefully the final coincident event list, to a more manageable level. This paper presents one such veto method that can be used to veto certain classes of transient events detected in the output data stream of GEO 600. The method uses events detected in the null-stream output of GEO 600 (which contains, in principle, no gravitational wave signal) with a threshold to veto events detected in the main strain output. We show that, for the certain types of signals tested, the method is very robust, delivering high efficiency for a very low false-veto rate. In particular, it is shown that when applied to real detector data, the method is able to strongly veto a certain type of events which appear around 370 Hz in the detector output.